Energy 34 (2009) 11131121

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

Local entropy generation for saturated two-phase ow

Remi Revellin a, *, Stephane Lips a, Sameer Khandekar b, Jocelyn Bonjour a
a
b

Universite de Lyon, CNRS INSA-Lyon, CETHIL, UMR5008, F-69621, Villeurbanne, France, Universite Lyon 1, F-69622, France
Department of Mechanical Engineering, Indian Institute of Technology Kanpur, Kanpur 208016, India

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 18 September 2008
Received in revised form
23 March 2009
Accepted 28 March 2009
Available online 22 May 2009

This paper addresses the estimation of local entropy generation rate for diabatic saturated two-phase
ow of a pure uid. Two different approaches have been adopted for this thermodynamic characterization: the separated ow model using the classical vapor ow quality, and the mixture model, using the
thermodynamic vapor quality. Based on these two models, two distinct expressions for the local entropy
generation have been proposed. The analysis explicitly shows the contribution of heat transfer and
pressure drop respectively to the local entropy generation. The contribution due to phase-change process
is also determined using the mixture model. The developed formulation is applied to analyze the
thermodynamic performance of enhanced heat transfer tubes under different conditions. It is shown that
enhanced tubes may be a relevant solution for reducing entropy generation at low mass velocities
whereas smooth tubes remain the best solution at higher ones.
2009 Elsevier Ltd. All rights reserved.

Keywords:
Entropy generation
Two-phase ow
Saturated ow boiling
Clapeyron equation

1. Introduction
In this paper, we focus our attention on the local entropy
generation during diabatic saturated two-phase ow. The primary
motivation of the study is to quantify local entropy generation under
such ow conditions. In the literature, efforts to evaluate numerous
heat transfer related processes in terms of entropy generation have
been promoted. Bejan [1] studied entropy generation through heat
and uid ow of a single-phase uid. He proposed an entropy
0
0
generation number given by NS S_ gen;enhanced =S_ gen;smooth. More
recently, Bejan [2] made a review on the thermodynamic optimization (or entropy generation minimization) of ow geometry in
engineering ow systems. It was observed that design improvements are registered by spreading the imperfections (e.g. ow
resistances) through the system. An expression was also proposed
for the entropy generation during a diabatic single-phase ow
indicating that irreversibilities come from two main causes: heat
transfer and pressure drops. Based on the entropy generation
theorem, Zimparov [3] has proposed equations for enhanced heat
transfer surfaces to include the effect of uid temperature variation
along the length of a tubular heat exchanger during single-phase
ow. This general evaluation criteria add new information to Bejans
entropy generation minimization method, assessing the two
objectives simultaneously. They may help to identify inappropriate
enhanced surfaces, in terms of irreversibilities, thereby assisting in
designing better heat transfer equipment. The article ended with
* Corresponding author. Tel.: 33 4 72 43 72 31; fax: 33 4 72 43 88 11.
E-mail address: remi.revellin@insa-lyon.fr (R. Revellin).
0360-5442/$ see front matter 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.energy.2009.03.014

the estimation of heat transfer and uid friction characteristics of

ten spirally corrugated tubes. Thermodynamic optimization
method of tree-shaped ow geometries during single-phase ow
was also proposed by Zimparov et al. [4]. They optimized the
performance of several classes of simple systems consisting of Tand Y-shaped assemblies of ducts, channels and streams. They
assumed a laminar and fully developed ow and developed
analytical equations for determining the entropy generation in
these geometries. Abbassi [5] has performed an entropy generation
analysis in a uniformly heated microchannel heat sink (MCHS) by
using the porous medium approach based on extended Darcy
equation for uid ow, and two-equation model for heat transfer.
They investigated the effect of inuential parameters such as,
channel aspect ratio, thermal conductivity ratio of the uid and the
solid and effective porosity of the MCHS on the total entropy
generation. Lior et al. [6] has reviewed the methods developed for
analyzing the space and time dependent exergy and irreversibility
elds in transport processes. They also emphasized the fact that
only few articles suggest keys to process improvements. Sahiti et al.
[7] developed an optimization model on the basis of the entropy
generation minimization for different double-pipe heat exchanger
(air/water) ow lengths and different length of circumferential pin
ns. Their conclusions are derived on the basis of the behavior of
entropy generation number as a function of Reynolds number.
Owing to the fact that the spatial distribution of the air-side heat
transfer coefcient is difcult to establish accurately, their formulation is based on the overall heat transfer coefcient, and cannot
therefore be termed as a local expression. Haddad et al. [8] studied
the entropy generation due to laminar forced convection in the

1114

R. Revellin et al. / Energy 34 (2009) 11131121

Nomenclature
A
Ar
Be
cp
cx
dz
D
Ec
G
h
L
m
_
m
NS
p
P
q
Q_
s
0
dS_
dt
T
DT

cross section, m2
area ratio between enhanced and smooth tubes based
on the internal diameter
0
0
Bejan number Be dS_ heattransfer =dS_
specic heat capacity at constant pressure, J/kg K
specic heat capacity at constant vapor quality, J/kg K
element of discretization, m
tube diameter, m
Eckert number Ec U 2 =cp DT
mass velocity, kg/m2 s
specic enthalpy, J/kg
length, m
mass, kg
mass ow rate, kg/s
entropy generation number
pressure, Pa
perimeter, m
heat ux, W/m2
heat rate, W
specic entropy, J/kg K
entropy generation rate per unit length, W/K m
characteristic time, s
temperature, K
wall superheat, K

entrance region of a concentric cylindrical annulus. They found that

the entropy generation was inversely proportional to both Reynolds
number and the dimensionless entrance temperature. The results
also showed that increasing Eckert number and/or the radius ratio
will increase the entropy generation.
In addition to these single-phase studies, some authors have
also developed theories on entropy generation for two-phase ow.
Collado [9] has established an expression for the entropy-based
boiling curve for subcooled ow boiling. The work is based on the
classical 1-D entropy formulation for a steady boiling ow.
However, in this article, the pressure drop term has been neglected.
This might be inappropriate for the case of ow boiling in mini or
microchannels, for which, high friction losses are expected. Besides,
the specic contributions, due to heat transfer, pressure drop and
phase-change, to the entropy generation, have not been explicitly
expressed. More recently, a mathematical model was presented by
Saechan and Wongwises [10] for determining optimal conguration of plate nned tube condensers (air-refrigerant type). The
analysis was based on the minimum entropy generation of air and
refrigerant. The equations were based on the overall heat transfer
coefcient. The entropy generation number, NS, was used for optimizing their architecture (e.g. number of tube rows, n spacing,
tube diameter, etc.). They concluded their article by saying that
most of the entropy generation occurred on the air side. Another
theoretical model for two-phase ow characterization was developed by Bilicki et al. [11] by the method of irreversible thermodynamics. They considered only particulate uid ows (i.e. not
annular, stratied nor churn ow regimes) and subdivided their
uid volume into a large but nite number of cells. They wrote the
conservation laws for two-phase ow. Their model is however
complicated and the authors needed some appropriate approximations or even transformations to use their model. Finally, they
restricted their model to a 1-D homogeneous model but they were
able to predict some experimental data. They also obtained
expressions for rate of local entropy generation per unit volume.
Shiba and Bejan [12] optimized the global performance of

U
v
x
z

velocity, m/s
specic volume, kg/m3
vapor quality
longitudinal abscissa, m

Greek symbols
b
parameter
G
slip ratio (Uv/Ul)
r
density, kg/m3
Subscripts
enhanced refers to enhanced tubes
gen
generated
in
inlet
l
liquid
lv
liquid to vapor
heattransfer heat transfer
m
mixture
min
minimum
t
total
tp
two-phase
sat
saturation
smooth refers to smooth tubes
v
vapor
w
wall

a counterow heat exchanger serving as a condenser in a vaporcompression-cycle refrigeration system by minimizing the total
entropy generation rate. Their numerical results showed how the
optimal conguration responded to changes in specied external
parameters such as refrigeration load, fan efciency, volume and
weight. This interesting work was however devoted to the system
and not the two-phase ow itself, which is the objective of the
present paper. Vargas and Bejan [13] considered the fundamental
problem of thermodynamically matching the two streams, one hot
and the other cold, so that the irreversibility generated by their
thermal interaction is minimum. Each stream experienced a change
of phase. The authors showed that the optimum was marked by an
optimal ratio between the stream mass ow rates, and an optimal
ratio between the two heat exchanger sizes when the total heat
transfer area was xed. This study was also devoted to the system
approach and not the local aspect of two-phase ow.
From the literature review, we highlight the fact that, while
quite some attempts have been made in the past to estimate
entropy generation in single-phase ows, almost no study has been
performed on the estimation of local entropy generation for a diabatic saturated two-phase ow. Moreover, most of the earlier
studies are dedicated to single-phase ow or, at best, to global
space-time averaged analysis of two-phase ow on a system level.
To begin with, we will redene some thermodynamic relations for
1-D diabatic saturated two-phase ow of a pure uid based on two
different approaches: the separated ow model and the mixture
model. Thereafter, we will propose two new local entropy generation equations based on these two models respectively. Finally, we
will demonstrate the use of this formulation with a case study of
enhanced heat transfer tubes.
2. Energy and mass conservation for a diabatic saturated twophase ow
Two different approaches may be found in the literature for
modeling two-phase ow: the relatively simpler homogeneous

R. Revellin et al. / Energy 34 (2009) 11131121

model which assumes a slip ratio (G Uv/Ul) equal to unity (both

phases travel at the same velocity) and the more detailed separated
ow models. However, if we want to treat two-phase ow from
a more holistic viewpoint, both these conventional models have
inherent limitations and the use of the mixture model is necessary.
We will present the basic equations for the separated ow model
and for the mixture model in the next sections.
2.1. Separated ow model
Usually, the classical vapor quality (also called ow quality)
equation used for a diabatic two-phase ow is,

_v
m
_ vm
_l
m

(1)

Therefore, the two-phase enthalpy and two-phase entropy are

expressed as:

htp xhv 1  xhl

(2)

and

stp xsv 1  xsl

(3)

(4)

Assuming that the kinetic and gravity terms are much less than
the thermal energy component, rst law of thermodynamics gives:

dQ_
_t
m

 
d htp hlv dx xdhv 1  xdhl

(5)

dhv Tv dsv yv dpv

dhl Tl dsl yl dpl

(6)

Assuming that Tv Tl Tsat and dpv dpl dp, neglecting

capillary contribution, we get

_t
m

2.2. Mixture model

The mixture theory presented in this paper was proposed by
Collado [14] and is based on the Reynolds transport theory [15]. In
this paper, we will go one step further from Collado [4]: we will not
necessarily assume a constant slip ratio (as was originally done in
Ref. [15]) and we will also explicitly take into account the pressure
drop contribution to ow irreversibilities. The reasoning of the
theory is quite simple: if the phases travel at different velocities,
they cannot cover the same distance during the same time. As
a result, for the same length of control volume, the time scale is
different for each phase: dtv refers to the characteristic time of the
vapor phase and dtl refers to the characteristic time of the liquid
phase. To take into account the fact that the temporal scales are
distinct for both the phases, a time scaling factor should be
included, which, as shown hereafter, is the slip ratio (G).
Consider a xed mass of uid mm composed of saturated vapor
and liquid phase, owing in a diabatic channel:

(11)

If we follow (i.e. we track) the same xed mass of uid through

time and space inside the pipe, we can divide all the terms by dtl:

mm
mv ml

dtl
dtl dtl

(12)

or

dtv mv ml
mm

dtl
dtl dtv dtl

Furthermore, we know that

dQ_

With this particular model, both phases travel at different

velocities. The distance covered by the vapor and the liquid is thus
different over the same time dt.

mm mv ml

The two-phase specic volume is dened as:

vtp xyv 1  xyl

1115

(13)

with

dtl
Uv

G
dtv
Ul

(14)

Finally, for the limit case dtl /0, we obtain

hlv dx xdsv 1  xdsl Tsat xyv 1  xyl dp

hlv dx cx;tp dTsat ytp dp

(7)

The heat capacity of the two-phase ow at constant vapor

quality is dened as:

cx;tp xcx;v 1  xcx;l

(8)

_m
m

dsl
dTsat x0
dsv
Tsat dT
sat x1

cx;l Tsat
cx;v

rm

(9)

Using the Clapeyron equation dp=dTsat hlv =Tsat ylv , Eq. (7)
becomes:

dx

dQ_
_ t hlv
m




ylv Tsat cx;tp hlv ytp

h2lv


dp

flashing effect

_ l rm Ul A
m

(15)

xm

rv

1  xm

rl

1

arv 1  arl

(16)

where a is the volumetric fraction of the vapor in the ow boiling

mixture, which can be measured with reasonable accuracy from Xor a-ray methods [16].
The mixture vapor quality corresponds to the thermodynamic
vapor quality and is expressed as:

(10)

_ t is the total mass

where dQ_ is the heat supplied to the uid and m
ow rate of the two-phase ow (constant).
In Eq. (10), the second term refers to the enthalpy rearrangement between the phases due to the spontaneous ashing effect of
the two-phase ow, which becomes signicant when the pressure
drop is high.

_ t is the total
where A is the cross section of the channel. Note that m
mass ow rate of the inlet uid, which usually is liquid. Therefore,
_ t must be that of the liquid.
the time reference associated with m
The mixture density is thermodynamically dened as:

with

_v
m

xm

_v
_v
m
m
mv

_
_
_
mv Gml
mm
mv ml

(17)

We may express the thermodynamic vapor quality as a function

of the classical vapor quality (ow quality) given by Eq. (1):

xm

x
x G1  x

(18)

1116

R. Revellin et al. / Energy 34 (2009) 11131121

Eq. (18) has been plotted in Fig. 1 for different slip ratios. As
expected, higher the slip ratio, greater is the deviation between
thermodynamic and conventional vapor quality; the former being
always lower than the latter for G > 1. For G 1, we obtain xm x.
The elementary derivative of the thermodynamic vapor quality
is given by the following relation:

dxm

G
2

x G1  x

dx 

x1  x
x G1  x2

dG

(19)

The mixture enthalpy and entropy (still tracking the same

constant mass) are thermodynamically expressed as:

hm xm hv 1  xm h1

(20)

Fig. 2. Schematic of a diabatic saturated two-phase ow in a channel.

temperature which is assumed to be the temperature of the system

boundary, as a rst approximation. Furthermore, we know that
_ t dstp dm
_ v sv m
_ l sl , thus we get:
dhtp Tsatdstp vtpdp and m

and
0

sm xm sv 1  xm sl

(21)

The equations obtained in this section (mixture model) are

applicable irrespective of the equilibrium state of the ow, i.e. the
liquid and vapor temperatures may exhibit different values, i.e. hv,
sv to be calculated at Tv and hl, sl to be calculated at Tl (Tv may or may
not be equal to Tl). However, in this study, local thermodynamic
equilibrium is assumed in each discretized cell. As a consequence,
we will take all the properties at Tv Tl Tsat. Note also, that for
G 1, the mixture model and the separated ow models are
equivalent.

dS_ dz

_ t vtp
_t
m
dQ_
m
dhtp 
dp 
Tsat
Tsat
Tw

(23)

Combining the second law constraint with the energy equations,

i.e. Eqs. (23) and (5), it is simplied as:

dQ_

dS_ dz

Tsat

dQ_
Tw

_ t ytp
m
dp
Tsat

(24)

For an adiabatic ow, Eq. (24) reduces to,

0

dS_



_ t ytp
m
dp

Tsat
dz

(25)

3. Local entropy generation for saturated two-phase ow

For a diabatic ow, we can write that Tw Tsat DT (DT being the
superheat of the wall), as a consequence, we get,

3.1. Separated ow model

#
"


_ t ytp
m
dQ_
DT
0
dp
_


dS
2 1 DT=T
Tsat
dz
dz Tsat
sat

Assume a saturated two-phase ow in a channel owing at

_ t as shown in Fig. 2. According to the
a certain mass ow rate m
second law of thermodynamics applied to a control volume of
length dz, we may write the following relation:


 dQ_
0
_ v sv m
_ l sl 
dS_ dz d m
Tw

(22)

where dS_ is the entropy generation rate per unit length, dQ_ is the
heat rate applied to the control volume and Tw is the wall

Slip ratio = 1
Slip ratio = 2
Slip ratio = 5
Slip ratio = 10

Thermodynamic vapor quality []

0.8

In
general,
DT/Tsat < < 1. Besides, we know that
q hDT dQ_ =dzP (assuming a constant perimeter P of the
channel with respect to z). Thus, this yields,
0
dS_
|{z}

entropy generation

q2 P
2
hTsat
|{z}

heat transfer contribution



_ t ytp
m
dp

dz
Tsat
|{z}

pressure drop contribution

(27)
where q is the heat ux and h is the heat transfer coefcient at the
tube wall.
Eq. (27) can be applied to any diabatic two-phase ow. The term
dp=dz indirectly takes into account part of the slip ratio effect.
However, the equation is not complete because it does not account
for the entire phase-change contribution.

1
0.9

(26)

0.7
0.6

3.2. Mixture model

0.5

The separated ow model and the mixture model are different,

as has been discussed earlier. We now introduce a link between
both models by dening a factor F:

0.4
0.3
0.2

dhtp

0.1
0

0.2

0.4

0.6

0.8

Classical vapor quality []

Fig. 1. Thermodynamic vapor quality as a function of the classical vapor quality.

dQ_
_
m

Fdhm

(28)

According to Fig. 1 and assuming that dhtp z hlvdx and

dhm z hlvdxm, it is easy to see that:
a) at low x and as G > 1, dx > dxm 0 dhtp > dhm, 0 F > 1
b) at high x and as G > 1, dx < dxm 0 dhtp < dhm, 0 F < 1

R. Revellin et al. / Energy 34 (2009) 11131121

As a result, from (b) we see that using F G (as has been done in
Collados theory) in this study is inappropriate. However, there
exists a relation between the vapor ow quality and the thermodynamic vapor quality, which is given by Eqs. (18) and (19).
Furthermore, the vapor ow quality can be expressed using
a simple energy balance (Eq. (10)). As a result, it is possible to
formulate an alternative and more comprehensive energy balance
for the mixture model, which would link the heat ux, the mass
ow rate and the mixture enthalpy. This energy balance formulation is described below.

1117

dQ_

_ t dhm
m



cx;m  bcx;tp ylv Tsat hlv ym  bytp
_ t dp
b
m
hlv dQ_

1
_ t hlv b
x1  xm
dG

GdQ_

(37)

or

Ghlv ddzQ






_
_ t dp  x1  xm
_ t h2 bdG
bGhlv ddzQ  cx;m  bcx;tp ylv Tsat hlv ym  bytp Gm
dz
lv dz

Assume a saturated two-phase ow mixture owing in

a channel as shown in Fig. 2. Starting with Eq. (10), we can write:

dQ_

hlv dx

_t
m

ylv Tsat cx;tp hlv ytp

hlv

dp

(29)

Combining with Eq. (19), we get

dQ_

hlv

_t
m


2
"
#
dxm x G1  x x1  xdG

_t
m

(39)

G
ylv Tsat cx;tp hlv ytp
hlv

As a result, for an adiabatic two-phase ow

dp

(30)
F

x1  xdG

b hlv dxm hlv b

bhlv ytp bcx;tp ylv Tsat

hlv

dp

(31)

with

The arising of the factor F dividing the heat rate is justied by

the former mass balance. As a matter of fact, the time scales of
the liquid and the vapor must be different because the phase
velocities are not equal. Therefore, F would act as a scaling factor
between the two different time scales associated with the twophases respectively. For an adiabatic two-phase ow, dhm 0,
and we get





cx;m  bcx;tp ylv Tsat hlv ym  bytp Gdp x1  xh2lv bdG

or

dQ_

(38)

(32)

x G1  x2

It is interesting to note that b 1 when G 1. Since

dhm hlv dxm cx;m dTsat ym dp

(40)

Note that, at low vapor ow qualities, the pressure drop and

phase-change terms are negligible compared to the heat transfer
term in the expression of F. Thus, we can say that F z G which is the
factor determined experimentally by Collado et al. [17] for such
conditions at low x. Furthermore, when G 1, we get F 1. Now it is
possible to write the expression of the entropy generation for the
two-phase ow mixture:
0

_ t dsm 
dS_ dz m
(33)

dQ_
Tw F

(41)

or

we nally get:

_t
_ t ym
dQ_
m
m
dhm 
dp 
Tsat
Tsat
Tw F

dhm

dS_ dz





cx;m  bcx;tp ylv Tsat hlv ym  bytp
dQ_
b

dp
_t
m
hlv


x1  xhlv b

dG

_ t dhm , and we get

By denition, dQ_ =F m

(34)

where

cx;m xm cx;v 1  xm cx;l

(35)

dQ_

dS_ dz

Tsat F

dsl
dT
 sat xm x0
dsv
Tsat dT
sat x x1
m

dQ_
Tw F

_ t ym
m
dp
Tsat

(43)

For the same reasons explained in the previous section, we can

write

dS_
|{z}

and

cx;l Tsat

(42)

entropy generation

q2 P
2 F
hTsat
|{z
}

heat transfer contribution



_ t ym
m
dp

dz
Tsat
|{z}

pressure drop contribution

(36)

(44)

Based on Eq. (34), we can introduce a factor F which gives the

relation between dQ_ and dhm.

which is the entropy generation rate (with the time scale of the
liquid) per unit length. Again, we can determine the heat transfer
and pressure drop contributions to the entropy generation. When

cx;v

1118

R. Revellin et al. / Energy 34 (2009) 11131121

G 1, Eq. (44) is equivalent to Eq. (27), as expected. This new

expression is explicit and highlights the effect of the slip ratio on
the entropy generation, not only in the dp term but also in F. The
term with dG in the expression of F (Eq. (38)) corresponds to the
phase-change contribution which was not taken into account in Eq.
(27). In that sense, this new expression based on the mixture model
is more comprehensive than expression (27).
4. Applications
We now highlight the application of Eqs. (27) and (44) in
practical saturated two-phase ow boiling congurations. Usually,
it is accepted that the use of enhanced tubes for ow boiling is
a good way to increase heat transfer compared to smooth tubes.
Meanwhile, the pressure drops are much higher with enhanced
tubes than with smooth tubes. So, one of the important questions
is: Are enhanced tubes more suitable under all ow conditions?
Relations (27) and (44) have been applied to a smooth tube of
10 mm inside diameter and an enhanced tube of the same equivalent diameter, both having a length of 3 m. The heat transfer coefcient for the smooth tube has been calculated with the ow pattern
based model by Wojtan et al. [18]. The pressure drops have been
determined using the ow pattern based model by Moreno Quiben

0.1

245 kg/m2s
0.05

150

200

The results of the simulations using the separated ow model

are shown in Fig. 3(a). The entropy generation has been plotted
versus the mass velocity for R-134a at a saturation temperature of
10  C for a heat ux of 10 kW/m2. It clearly appears that for a mass
velocity less than 245 kg/m2 s, the use of the enhanced tube is more
relevant because the entropy generation is lower. Actually, at low
mass velocities, the benet of the enhanced tubes comes from the
heat transfer contribution whereas the pressure drop contribution
remains small. At mass velocities greater than 245 kg/m2s, the use
of enhanced tubes is no more advantageous because the pressure
drop penalty is too high.

Plain tube
Enhanced tube

0
100

4.1. Results

Entropy generation [W/K]

Entropy generation [W/K]

and Thome [19] for smooth tubes. Regarding the enhanced tube, the
model by Koyama et al. [20] has been used for calculating the heat
transfer coefcient of the micron tube. The pressure drops have
been determined using the correlation by Kuo and Wang [21]. The
area ratio (Ar) between the enhanced and the smooth tubes is
arbitrarily set at 1.5. The void fraction has been determined using the
Steiner [22] version of the Rouhani and Axelsson [23] drift ux void
fraction model. Tube orientation is horizontal. As a result, the
gravitational pressure drop is non-existent. The acceleration and
frictional pressure drop have been taken into account.

250

300

350

400

Plain tube
Enhanced tube

0.1

0.05
235 kg/m2s

0
100

450

150

Mass velocity [kg/m2s]

b
Entropy generation number []

Entropy generation number []

300

350

400

450

2
1.8

1.8
NS>1 favorable for smooth tubes

1.4

NS

1.2
1
0.8
0.6
NS<1 favorable for enhanced tubes

0.4

1.6 NS>1 favorable for smooth tubes

1.4
NS

1.2
1
0.8
0.6

NS<1 favorable for enhanced tubes

0.4
0.2

0.2
0
100

250

Mass velocity [kg/m2s]

1.6

200

150

200

250

300

350

400

450

Mass velocity [kg/m2s]

Fig. 3. Simulations using the separated ow model for R-134a, Tsat 10  C, xin 0.3,
q 10 kW/m2, D 10 mm and L 3 m. (a) Entropy generation for smooth and
enhanced tubes as a function of the mass velocity. (b) Entropy generation number as
a function of the mass velocity.

0
100

150

200

250

300

350

400

450

Mass velocity [kg/m2s]

Fig. 4. Simulations using the mixture model for R-134a, Tsat 10  C, xin 0.3,
q 10 kW/m2, D 10 mm and L 3 m. (a) Entropy generation for smooth and
enhanced tubes as a function of the mass velocity. (b) Entropy generation number as
a function of the mass velocity.

R. Revellin et al. / Energy 34 (2009) 11131121

- for G < 170 kg/m2 s, slug and stratied wavy, stratied wavy
and dryout,
- for G > 170 kg/m2 s, slug, intermittent, annular, dryout and mist
ow.

500
450

Mass velocity [kg/m2s]

400

350

Describing each ow pattern based model for heat transfer and

pressure drop is not the purpose of this work. However, it is
important to understand that entropy generation is directly related
to the ow congurations and therefore it is important to study the
ow regimes and their transitions.
Heat transfer and pressure drop contributions are depicted in
Fig. 6 using the Bejan number which is dened as:

300
250

Slug

200
150

Slug+SW

100

Be

SW
50
S
0

0.2

0.4

0.6

0.8

Vapor flow quality []

Fig. 5. Flow pattern map for R-134a, Tsat 10  C, q 10 kW/m2, D 10 mm and
G 250 kg/m2s (Wojtan et al. [24]): S: Stratied; SW: Stratied wavy; I: Intermittent;
A: Annular; D: Dryout; and M: Mist ow.

Another way to present the results could be the calculation of

the entropy generation number. In our case, NS is dened as:
0

NS

1119

dS_ enhanced
0
dS_

(45)

dS_ heattransfer
0
dS_

The Bejan number shows the importance of heat transfer

contribution to the total entropy generation. We can see that at low
mass velocity, both models give the same results. However, at high
mass velocities, the pressure drop contribution for the mixture
model is higher due to the phase-change contribution. As a consequence, the entropy generation per unit length will be larger. We
also clearly see that at low mass velocity, entropy generation is
mainly due to the heat transfer whereas at high mass velocity,
entropy generation is mostly created by pressure drop, which is

D = 9 mm
D = 10 mm
D = 12 mm
D = 14 mm

Entropy generation [W/K]

smooth

Fig. 3b shows the results for this entropy generation number

using the separated ow model. The same conclusions as before
may be drawn.
Regarding the mixture model, simulations are presented in
Fig. 4 for the same ow and geometric conditions as those of Fig. 3.
It is clear that for a mass velocity less than 235 kg/m2 s, the use of
the enhanced tubes is more relevant because the entropy generation is lower.
Entropy generation is directly related to ow patterns. Indeed,
heat transfer coefcients or pressure drops are based on the
different ow patterns encountered during evaporation and are
shown in Fig. 5:

(46)

0.1

0.05

0
100

150

200

250

300

350

400

450

400

450

Mass velocity [kg/m2s]

0.9

0.9

0.8

0.8

Bejan number Be []

Bejan number Be []

0.7
0.6
0.5
0.4
0.3

0.7
0.6
0.5
0.4
0.3
0.2

0.2
0.1
0
100

Plain tube, mixture model

Enhanced tube, mixture model
Plain tube, separated flow model
Enhanced tube, separated flow model
150

200

250

300

0.1
0
100
350

400

450

Mass velocity [kg/m2s]

Fig. 6. Bejan number as a function of the mass velocity for both tubes and the two
models.

D = 9 mm
D = 10 mm
D = 12 mm
D = 14 mm
150

200

250

300

350

Mass velocity [kg/m2s]

Fig. 7. Inuence of the diameter using the separated ow model for R-134a,
Tsat 10  C, xin 0.3, q 10 kW/m2 and L 3 m (enhanced tube). (a) Entropy generation versus mass velocity. (b) Bejan Number versus mass velocity.

R. Revellin et al. / Energy 34 (2009) 11131121

higher for the enhanced tube. Finally, it follows that, enhanced

tubes are a relevant solution for reducing entropy generation at low
mass velocities whereas smooth tubes are better at higher ones.
The inuence of the diameter on the entropy generation has been
plotted in Figs. 7(a) and 8(a) for the enhanced tube. As it can be
observed, a relevant solution for reducing the entropy generation is
to decrease the diameter of the channel. This would promote the
heat transfer coefcient much more than the pressure drop, yielding
a decrease of the entropy generation. This increase of the heat
transfer contribution is shown in Figs. 7(b) and 8(b). The Bejan
number decreases when the diameter decreases. Investigations of
diabatic saturated two-phase ow in mini-channels is thus relevant.
Beyond the meaning of these results, one should be aware that
minimization of irreversibilities of an inherently irreversible
process should also be judged at the system level. Therefore, this
work should be extended to the system level in a second step.
Nevertheless, these sample applications show the simplicity as well
as the capability of the equations developed in this paper.

Entropy generation [W/K]

D = 9 mm
D = 10 mm
D = 12 mm
D = 14 mm
0.1

0.1

0.05

150

200

250

300

350

400

450

400

450

Mass velocity [kg/m2s]

Plain tube
Enhanced tube

0.1
Uncertainties

0.05

0
100

150

200

250

300

350

Mass velocity [kg/m2s]

0.05

Fig. 9. Uncertainty calculations for R-134a, Tsat 10  C, xin 0.3, q 10 kW/m2,

D 10 mm and L 3 m, using an uncertainty of 20% on the heat transfer coefcient
and the pressure drop. (a) Separated ow model. (b) Mixture model.

0
100

150

200

250

300

Mass velocity

350

400

450

[kg/m2s]

1
0.9
0.8

Bejan number Be []

Uncertainties

0
100

Entropy generation [W/K]

Uncertainty calculations related to the entropy generation are

discussed hereafter. The respective equations are iteratively solved
at the local saturation temperature corresponding to each

Plain tube
Enhanced tube

4.2. Uncertainty calculations

a
Entropy generation [W/K]

1120

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
100

D = 9 mm
D = 10 mm
D = 12 mm
D = 14 mm
150

200

discretized cell, with a convergence criteria of 0.001 K. The local

thermo-physical properties at the corresponding saturation
temperature are successively used. The tube is discretized into
5000 elementary control volumes (which is sufcient for grid
independent results) and typically, the solution converges in 34
cycles. The relevant uncertainty in the estimation of entropy
generation comes only from the models or correlations used for the
heat transfer coefcient and the pressure drop. These models are
empirical methods, which usually exhibit an error of 20%. This
typical uncertainty on the prediction methods has been simulated
in Fig. 9. We see that, the uncertainty is lower for the mixture
model than for the separated model. The uncertainty is really
dependent on the models used for the pressure drop or the heat
transfer coefcient calculation. The prediction methods used here
are believed to be the current best methods available in the open
literature.
5. Conclusions

250

300

350

400

450

Mass velocity [kg/m2s]

Fig. 8. Inuence of the diameter using the mixture model for R-134a, Tsat 10  C,
xin 0.3, q 10 kW/m2 and L 3 m (enhanced tube). (a) Entropy generation versus
mass velocity. (b) Bejan Number versus mass velocity.

A diabatic saturated two-phase ow of a pure uid has been

thermodynamically characterized based on two different
approaches: the separated ow model using the classical ow
quality and the mixture model using the thermodynamic vapor
quality. Two expressions for the local entropy generation have been

R. Revellin et al. / Energy 34 (2009) 11131121

proposed in this paper based on these two models. These equations

show the heat transfer, pressure drop and phase-change (for the
mixture model only) contributions to entropy generation. The
developed expressions of the entropy generation also present many
advantages, namely:
- They highlight the importance of accurately studying the heat
transfer coefcients, the pressure drops and the void fraction,
which allows the calculation of the slip ratio.
- They are local expressions which can be used in many geometries, e.g. smooth tubes, enhanced tubes, microchannels, etc.
- They can be used for evaporation or condensation.
- They propose an objective quantity for characterizing the
geometries (using the same uid) or choosing the uids (using
the same geometry).
Using these relations, it is shown that enhanced tubes may be
a relevant solution for reducing entropy generation at low mass
velocities whereas smooth tubes remain the best solution at higher
ones. Besides, under certain operating conditions, decreasing the
channel diameter would promote the heat transfer and reduce the
entropy generation.

References
[1] Bejan A. Entropy generation through heat and uid ow. New York: Wiley;
1982.
[2] Bejan A. Thermodynamic optimization of geometry in engineering ow
systems. Exergy Int J 2001;4:26977.
[3] Zimparov VD. Extended performance evaluation criteria for enhanced heat
transfer surfaces: heat transfer through ducts with constant heat ux. Int J
Heat Mass Transfer 2001;44:16980.
[4] Zimparov VD, da Silva AK, Bejan A. Thermodynamic optimization of treeshaped ow geometries. Int J Heat Mass Transfer 2006;49:161930.
[5] Abbassi H. Entropy generation analysis in a uniformly heated microchannel
heat sink. Energy 2007;32:193247.
[6] Lior N, Sarmiento-Darkin W, Al-Sharqawi HS. The exergy elds in transport
processes: their calculation and use. Energy 2006;31:55378.

1121

[7] Sahiti N, Krasniqi F, Fejzullahu Xh, Bunjaku J, Muriqi A. Entropy generation

minimization of a double-pipe pin n heat exchanger. Appl Therm Eng
2008;28:233744.
[8] Haddad OM, Alkam MK, Khasawneh MT. Entropy generation due to laminar
forced convection in the entrance region of a concentric annulus. Energy
2004;29:3555.
[9] Collado FJ. The law of stable equilibrium and the entropy-based boiling curve
for ow boiling. Energy 2005;30:80719.
[10] Saechan P, Wongwises S. Optimal conguration of cross ow plate nned tube
condenser based on the second law of thermodynamics. Int J Therm Sci 2008;
doi:10.1016/j.ijthermalsci.2007.12.009.
[11] Bilicki Z, Giot M, Kwidzinski R. Fundamentals of two-phase ow by the
method of irreversible thermodynamics. Int J Multiphase Flow 2002;28:
19832005.
[12] Shiba T, Bejan A. Thermodynamic optimization of geometric structure in the
counterow heat exchanger for an environmental control system. Energy
2001;26:493511.
[13] Vargas JVC, Bejan A. Thermodynamic optimization of the match between two
streams with phase change. Energy 2000;25:1533.
[14] Collado FJ. The entropy balance for boiling ow. Fusion Eng Design 2001;
5657:199203.
[15] Collado FJ. Reynolds transport theorem for a two-phase ow. Physics Lett
2004;90:024101.
[16] Collado FJ, Monne C, Pascau A. A new heat balance for ow boiling. AIChE J
2007;53:212330.
[17] Collado FJ, Monne C, Pascau A, Fuster D, Medrano A. Thermodynamics of void
fraction in saturated ow boiling. J Heat Transfer 2006;128:6115.
[18] Wojtan L, Ursenbacher T, Thome JR. Investigation of ow boiling in horizontal
tubes: part II development of a new heat transfer model for stratied-wavy,
dryout and mist ow regimes. Int J Heat Mass Transfer 2005;48:297085.
[19] Moreno Quiben J, Thome JR. Flow pattern based two-phase frictional pressure
drop model for horizontal tubes, part II: new phenomenological model. Int J
Heat Fluid Flow 2007;28:106072.
[20] Koyama S, Yu J, Momoki S, Fujii T, Honda H. Forced convective ow boiling
heat transfer of pure refrigerants inside a horizontal micron tube. In: Taylor
& Francis, editor. Proceedings of the convective ow boiling; 1995. p. 13742.
Alberta.
[21] Kuo C-S, Wang C-C. Horizontal ow boiling of R-22 and R-407C in a 9.52 mm
micron tube. Appl Therm Eng 1996;16:71931.
[22] Steiner D. Verein Deutscher Ingenieure VDI-Warmeatlas (VDI Heat Atlas),
chapter HBB, VDI-Gesellshaft Verfahrenstechnik und Chemieingenieurwesen
(GVC). Dusseldorf; 1993.
[23] Rouhani SZ, Axelsson E. Calculation of volume void fraction in subcooled and
quality region. Int J Heat Mass Transfer 1970;13:38393.
[24] Wojtan L, Ursenbacher T, Thome JR. Investigation of ow boiling in horizontal
tubes: part I a new diabatic two-phase ow pattern map. Int J Heat Mass
Transfer 2005;48:295569.